- 博客(55)
- 收藏
- 关注
RSS订阅原创 Detecting Everything in the Open World: Towards Universal Object Detection
在 Open World 中检测一切：面向通用目标检测。
2023-03-27 20:38:18
1757
原创 CurriculumNet: Weakly Supervised Learning from Large-Scale Web Images
1、在大规模弱监督网络网络图像上训练深度神经网络，图像为关键词索引的、没有任何人工注释的互联网图像，通过利用课程学习来制定学习策略以有效处理大量噪声标签和数据不均衡问题2、设计新的学习课程：通过在特征空间使用数据的分布密度来测量数据的复杂性，并以无监督的方式对复杂性进行排序，允许通过直接搜索高噪声标签，以实施有效的课程学习策略3、目的：提供一种能够有效处理大量噪声标签和数据不平衡的解决方案，通过利用课程学习来开发一种简单但高效的训练策略，通过利用高噪声标签来提高标准深度网络的模型泛化和整体能力。
2023-01-08 23:09:56
289
原创 Curriculum Learning
1、训练策略形式化，显著提高泛化能力，更快的收敛速度2、假设：精心选择的课程策略可以作为一种连续方法（continuation method）——能够帮助找到更好的非凸训练准则的局部最小值3、实验证明，课程策略类似于正则器，增益效果在测试集上更为明显，可以加速训练收敛到全局最小值4、
2023-01-08 12:12:14
214
原创 Multi-Lable 数据集
The MS COCO (Microsoft Common Objects in Context) dataset is a large-scale object detection, segmentation, key-point detection, and captioning dataset. The dataset consists of 328K images.Annotations: The dataset has annotations forThe first version of MS
2023-01-05 10:22:54
502
原创 Global Meets Local: Effective Multi-Label Image Classification via Category-Aware Weak Supervision
Global Meets Local: Effective Multi-Label Image Classification via Category-Aware Weak Supervision, 2022
2022-12-19 21:58:38
342
1
原创 Learning Disentangled Label Representations for Multi-label Classification
Learning Disentangled Label Representations for Multi-label Classification，2022
2022-12-19 21:41:42
685
1
原创 FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence
FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence, NIPS, 2020
2022-12-19 20:27:55
786
原创 Semi-Supervised Classification with Graph Convolutional Networks
Semi-Supervised Classification with Graph Convolutional Networks, ICLR, 2017
2022-12-08 23:20:01
668
原创 [CVPR2022] Debiased Learning from Naturally Imbalanced Pseudo-Labels
Debiased Learning from Naturally Imbalanced Pseudo-Labels
2022-11-26 23:48:27
986
原创 [CVPR2022] DASO: Distribution-Aware Semantics-Oriented Pseudo-Label for Imbalanced Semi-Supervised
DASO: Distribution-Aware Semantics-Oriented Pseudo-Label for Imbalanced Semi-Supervised Learning
2022-11-26 22:19:11
724
原创 [CVPR2022] Cross-Model Pseudo-Labeling for Semi-Supervised Action Recognition
Cross-Model Pseudo-Labeling for Semi-Supervised Action Recognition
2022-11-26 20:52:46
870
原创 [CVPR2022] BoostMIS: Boosting Medical Image Semi-Supervised Learning With Adaptive Pseudo Labeling a
BoostMIS: Boosting Medical Image Semi-Supervised Learning With Adaptive Pseudo Labeling and Informative Active Annotation
2022-11-26 14:06:10
794
2
原创 [CVPR2022] Back to Reality: Weakly-Supervised 3D Object Detection With Shape-Guided Label Enhancemen
Back to Reality: Weakly-Supervised 3D Object Detection With Shape-Guided Label Enhancement
2022-11-25 22:49:02
310
原创 [CVPR2022] ADeLA: Automatic Dense Labeling With Attention for Viewpoint Shift in Semantic Segmentati
ADeLA: Automatic Dense Labeling With Attention for Viewpoint Shift in Semantic Segmentation
2022-11-25 22:24:03
258
原创 [CVPR2022] ACPL: Anti-curriculum Pseudo-labelling for Semi-supervised Medical Image Classification
ACPL: Anti-curriculum Pseudo-labelling for Semi-supervised Medical Image Classification
2022-11-24 00:46:30
1559
原创 [CVPR2022] A Dual Weighting Label Assignment Scheme for Object Detection
A Dual Weighting Label Assignment Scheme for Object Detection
2022-11-24 00:36:28
622
原创 pytorch加载预训练模型
加载预训练模型1/4 直接加载预训练模型import torchvison.models as modelsresnet50 = models.resnet50(pretrained=True)# 若只需要网络结构，不需要预训练模型的参数来初始化，则设置pretrained=False，或者把resnet复制到自己的目录下，新建个model文件夹，可以参考下面的猫狗大战入门算法入门：https://github.com/JackwithWilshere/Kaggle-Dogs_vs_Cats_PyT
2022-03-09 15:54:36
1826
原创 代码检查规则
java代码检查规则源文件规范文件名文件编码特殊字符源文件组织结构规范许可证或版权声明：应放在最开头package语句：单独一行不换行import语句唯一的顶层类代码结构规范花括号：K&R原则缩进与换行：4个空格缩进空行命名规范驼峰命名格式类的命名格式常量命名格式OOP规约所有的POJO类属性必须使用包装数据类型，禁止使用基本数据类型所有的覆写方法，必须加@Override注解Object的equals方法容易抛空指针异常，应使用常量或确定
2021-11-28 12:30:40
633
原创 质量保证意识
质量意识提高软件和服务的研发质量和研发效率项目管理三要素：质量、时间、成本质量保证与测试：质量保证应贯穿项目的整个过程bug修复：测试人员激活bug；开发人员收到激活bug后，进行修复，bug状态为处理中，修复完成后状态设置为解决状态。Open(激活)、In Process(处理中)、Resolved(解决)、Closed(关闭)、Reopen(重启)。内部+外部用户反馈渠道，bug提交格式：标题、问题描述、复现步骤、能否复现、问题来源、问题影响、补充描述。项目开发的整体路径：需求阶段、设计阶
2021-11-28 12:30:06
254
空空如也
空空如也
TA创建的收藏夹 TA关注的收藏夹
TA关注的人